GIGO

Surveys and Statistics

Lies, damned lies and statistics -- the phrase has been used so often that it seems trite, yet it retains a vital kernel of truth that we ignore at our peril. That's something to bear in mind as more and more statistics pour out in the run-up to the election.

Of particular concern is the disturbing development of "televoting" whereby television viewers are encouraged to pay to record their opinion on some topic deemed suitably controversial by TV programmers. The first such "telesurvey" was taken as part of a Counterpoint programme on the banning of discrimination against homosexuals. Some 10,000 people rang in to "vote" on whether the amendment should have been passed or not. Just under 60% disagreed with the legislation, a result opposite to the strong anti-discrimination sentiment expressed in national surveys.

I was appalled to hear Ian Fraser suggest that this "survey" contained meaningful information about New Zealanders' attitudes to gays. He challenged MP Lianne Dalziel to explain the apparent discrepancy between his televote and the national survey, as if they had equal validity. It was a relief to hear Ms Dalziel point out the difference between the sort of results one would expect from a self-selected group phoning in and those gained from a properly designed social survey.

It should have been obvious to Mr Fraser that the very act of inviting opinion will encourage response from the extremes, those with strongly held opinions. The hi-tech, push-button approach of televoting has as much relationship to real public opinion polling as sound bites have to reasoned political debate. Sadly, I doubt that such crucial differences are appreciated by the sort of people likely to cite such telesurveys as justification for whatever opinions they hold.

My own MP recently announced strong electorate support for one of her pet projects, based on the results of a mail-in survey. I was one of the respondents (and one of the supporters), but the thought of government through such "referenda" makes me extremely uneasy. Professional surveys have enough holes in their methodology to make me wary -- the thought of making national decisions on the sort of shonky, sham statistics we see in these seat-of-the-pants polls is, frankly, frightening.

Vicki Hyde is the editor of New Zealand Science Monthly.